State estimation scheme for multiple RF tags with an angled single antenna

This paper proposes a state estimation scheme for multiple RF tags with an angled antenna. As we employ an angles antenna, the system can make a difference in communication features with a single antenna in a detection area. Since the response information about each RF tag is discrete time-series sampling due to accessing multiple RF tags, the system use statistic values such as mean, variance, etc., for Received Signal Strength Indication (RSSI), phase angle values. Since the statistic values hide characteristic signatures of movements, we apply machine learning to capture the features of tags' signals. As a result, it can estimate three states: stay, forward, and backward direction as the RF tags' states. The evaluation results show the estimation accuracy with different combinations of learning parameters in various communication environments.